MIDAS: Proactive Traffic Control System for Diamond Interchanges. Viswanath Potluri 1 Pitu Mirchandani 1

Transcription

1 MIDAS: Proactive Traffic Control System for Diamond Interchanges Viswanath Potluri 1 Pitu Mirchandani 1 1 Arizona State University, School of Computing, Informatics and Decision Systems Engineering 2 Arizona State University, School of Sustainable Engineering, and the Built Environment

2 What is MIDAS? Managing Interacting Demand And Supply Demand Supply

3 A Note on Control Systems Classification of control systems Open Loop Closed Loop (defined by feedback path) Open Loop In open loop control systems, output is not fed-back to the input. So, the control action is independent of the desired output. Fixed time signal control. Closed Loop In closed loop control systems, output is fed back to the input. So, the control action is dependent on the desired output. Actuated signal control & adaptive traffic control systems.

4 Adaptive Control System

5 Open-loop vs Reactive vs Proactive Position Actual Trajectory Reactive Proactive Open Loop Time

6 General Proactive Control Architecture

7 Measurements Eulerian Measurements Data collected at a fixed point in space, also called point detectors. Inductive loop detectors, video detectors, etc. They give traffic counts and approximate vehicle speeds. Lagrangian Measurements Data collected from mobile detectors that move with flow of traffic. Cell phones, GPS-based locator, etc. They give travel times, vehicle trajectories, speeds, etc. MIDAS uses Lagrangian measurements.

8 MIDAS for Diamond Interchanges Signal control of two closely spaced intersections at diamond interchange faces following challenges Complicated traffic movements Phase overlaps Limited inter storage capacity for queued vehicles Fluctuating demand and heavy off-ramp traffic

9 MIDAS Prediction & Control; Arterial PREDICT Arrivals & Queues Control Algorithms TURNING RATIOS TRAVEL TIMES 0DISCHARGE RATES ESTIMATION MODELS DATA STREAMING (GPS) State of traffic network

10 MIDAS Signal Control Algorithm MIDAS employs efficient Dynamic Programming(DP) approach to optimize traffic movements at the intersection, at a lane level resolution. Determines optimal phase sequence & duration of phases. Totally cycle free control strategy. Employs forward recursion DP approach to solve and backward recursion to retrieve optimal phase schedule. Flexible enough to optimize user defined performance measure, like stops, delays and queues, etc. over a finite time horizon that rolls forward.

11 MIDAS Signal Control Algorithm Decision variable: xx jj (Phase duration of stage j) Stage: Phase j State variable: ss jj (time horizon with stage j) Incremental value of objective function: ff ss jj, xx jj Cumulative value of objective function: VV jj 1 ss jj 1 VV jj ss jj = min{ff ss jj, xx jj + VV jj 1 ss jj 1, xx jj XX jj ss jj ss jj 1 xx jj r 0 T DP Solution

12 Diamond Interchange DP Solution Example A B C D E F G A E A C G DP Solution Example phase duration Possible Phase Combination

13 VISSIM Network Simulation I-17 & 19th Ave., Phoenix, AZ

14 Evaluation RHODES A predecessor to MIDAS Optimal Fixed Time Control (OFTC) VISSIM stage based optimization algorithm Sequence of simulation runs performed to determine best signal program. Signal program is constructed by modifying green times of best & worst stage. The stage with the lowest maximum average delay is selected as the best stage. The stage with the highest maximum average delay is selected as the worst stage.

15 Performance Metrics Delay Average Average of all vehicle delays due to presence of signal controller in their path when compared to free flow, without any signal control. Total Delay Sum of all vehicle delays in network, in seconds. Stops Average Average number of stops made by a vehicle. Average Queue Length Average queue length at any given movement at the stop line of the interchange. Total Travel Times Sum of travel times of all vehicles in the network, in seconds.

16 Results Network Level Performance SC SimTime(s) TraficLoad DelayAvg(s) TotalDelay(s) TotalTravelTime(s) MIDAS RHODES OFTC Intersection Level Performance SC SimTime(s) TraficLoad DelayAvg(s) StopsAvg(s) TotalStops(s) AvgQLEN MIDAS RHODES OFTC

17 Average Delay vs Traffic Load Average Delay (Secs) midas rhodes oftc Traffic Load (Veh/Hr)

18 Average Stops vs Traffic Load Avg. Stops midas rhodes oftc Traffic Load (Veh/Hr)

19 Average Queue Length vs Traffic Load Avg. Queue Length midas rhodes oftc Traffic Load (Veh/Hr)

20 References US DOT, FHWA; Chapter 2, Traffic Detector Handbook: Third Edition Volume I Mirchandani P. and K.L. Head A Real-Time Traffic Signal Control System: Architecture, Algorithms, and Analysis, Transportation Research Part C, Brent A. Cain, ADOT. I-17 & 19 th AVE. Proactive traffic control of diamond interchange project. Gettman, D., Head, K.L., and Mirchandani, P.B., RHODES-ITMS Corridor Control Project, Final Report, FHWA-AZ99-462, Arizona Department of Transportation, May 1999 (208 pages) Head, K.L, S. Joshua, S. Shelby, D. Gettman, P. Mirchandani, RHODES-ITMS: Real-Time Traffic Signal Control at a Diamond Interchange, 77th TRB Annual Meeting, Washington DC.. January 1998 Paper no

21 Thank You